Attempt Liability and Moral Luck (Part Two)

Tweet

(Part One, Series Index)

This is the second part in my short series on attempt liability and moral luck. As explained in part one, attempt liability is a concept from criminal law. It holds that it is legitimate to hold someone criminally liable for an attempted crime as well as for a completed crime. This poses certain problems, which were highlighted the last time. Those problems are connected to the more general problem of moral luck, which is the problem of determining how to account for the role of luck in our moral assessments.

Part one looked at three basic arguments in favour of attempt liability. In this part, we’ll look at three arguments on the other side of the debate. In compiling this list of arguments I’ve taken inspiration from the following article:

K. Kessler “The Role of Luck in the Criminal Law” (1994) 142 University of Pennsylvania Law Review 2183

But I haven’t covered every argument mentioned in that article; I’ve just selected the three I think are most interesting.


1. The No-Harm Argument
The first argument against attempt liability focuses (like the harm prevention argument from part one) on the purpose of the criminal justice system. More specifically, it focuses on the purpose of a criminalisation. It argues that certain actions, results and states of affairs are criminalised because they are harmful — in other words, it argues that the nature of a crime is such that it is something that is inherently harmful. The problem the argument then raises is that attempts are not, by nature, harmful. Therefore, it follows that attempts are not crimes.

• (1) For any X (where X is an action, result or state of affairs), X is a crime if and only if X is a harm.

• (2) Attempts are not harmful.

• (3) Therefore, attempts are not crimes.

There are many problems with this argument. One problem is that (2) could be false. Although some attempts might not be harmful, it could well be that others are harmful. Part of the issue here is how we define “harm”. Harms could be defined broadly to include psychological harms, or narrowly to only include physical harms (covering harms to property as well). The choice could be crucial in deciding whether an attempt is harmful since an attempt could cause psychological harm even if it didn’t cause physical harm. Then there’s the issue of how we define “attempt” itself. In the criminal law, an attempt is typically defined as something that is “more than merely preparatory” to an offence. Now, it’s quite possible that an action that is more than merely preparatory to murder could cause serious harm. So, under the definition just given, an attempt might be harmful. The problem with that, however, is that so many of harmful acts that fall short of murder are themselves criminalised. So it’s likely that if someone harmed another, but didn’t kill them, a lesser offence such as assault would be substituted. The only reason for choosing attempted murder over these lesser offences is if it leads to a higher punishment.

The bigger problem with this argument is that it is a conceptual/definitional argument, not a normative one. It focuses on the properties that an event or state of affairs must have in order to count as a crime. But the concept “crime” is not some kind of metaphysical necessity, etched into the fabric of the universe; it is instead a social construct, something whose boundaries we can change if we wish.

So for this argument to work we’d need to change premise (1) so that it becomes normative not definitional. For example:

• (1*) For any X (where X is an action, result or state of affairs), X should be criminalised if and only if X is a harm.

But when we do this the harm-essentialist view of crime might seem less plausible. We could then ask the question: why shouldn’t increasing the risk of harm be enough for X to count as a crime? Is that not something our risk averse society should seek? This leads us back to the harm prevention argument covered in part one.

Finally, another problem with the argument is that it focuses on the justification of criminalisation, not on the justification of punishment. When it comes to attempt liability it is the latter, not the former, that is important. This is significant because, if we adopted a utilitarian view of punishment, avoiding the risk of harm might be enough to justify the existence attempt liability. (Note: it could well be that criminalisation and punishment are inextricably linked so that justifying the one necessarily justifies the other. That’s an argument some people have made, but it’s not something I’ll get into here).



2. The Moral Luck Argument
This next argument links us directly back to the problem of moral luck, which was introduced in part one. As I noted there, one of the major developments in the 20th century analysis of moral luck, was the realisation that luck plays an important role in our everyday moral judgments. In particular, there was the realisation that we already seem to believe that the presence of luck should alter our moral judgments. And that our intuitive reactions and social practices reveal this to be true.

Consider the following example, drawn from a previous post:

Conference Organisation (1): You take responsibility for organising a conference. You invite the keynote speakers, send out the call for papers, book the venue, arrange for travel and so on. Everything is going well up until the day of the conference. Unfortunately, on that day, a freak snowstorm hits. No one can travel to the conference and the event has to be called off.

Conference Organisation (2): Same as (1) except that there’s no snowstorm. The conference is a resounding success.

Luck clearly separates these two cases. The conference organiser in the first example was a “victim” of what we might call bad luck, whereas in the second they are “victims” of good luck. Yet, when it comes to assessing both individuals, it is highly likely that the person in the second case will be rewarded for the success of their conference, while the person in the first case won’t be. But why is that? They both did the exact same things, the only difference between them is the kind of luck they were exposed to. Why should one be rewarded and the other not? The only way to explain this practice is to suppose that luck plays an important part in how we morally assess agents.

But this is just to point out how things actually are. How do we translate an observation about what is the case into an argument about what should be the case? After all, just because we do react this way, doesn’t mean we should. What will decide the matter is the weight we attach to intuitive reactions in our moral reasoning. If we attach a high weight to intuitions, then we might agree that luck should play a part in our moral reasoning. Hence, it would then seem to follow, that attempt liability is not (entirely) legitimate. On the other hand, if we attach more weight to principles (such as the control principle of responsibility) we might lean in the other direction.


3. The Deterrence Argument
People who read part one might be surprised to see deterrence cropping up again since in part one a deterrence argument was used to support the legitimacy of attempt liability. How can we now turn around and say that deterrence-based considerations support the other side of the debate? Very easily, actually. Deterrence is a tricky concept, its applicability depends crucially on the assumptions we make about human reactions to incentives. If those assumptions are wrong, or if there is some doubt about them, it is quite possible for a deterrence-based argument to work both ways.

So how does the deterrence argument against attempt liability work? Here’s a suggestion:

• (1) In order for an agent to be deterred from doing X (where X is an option the agent can exercise), the overall utility of X must be less than the overall utility of ~X.

• (2) If attempted offences are punished in the same way as completed offences, then the overall utility of an attempt is not more than (and possibly less than) the overall utility of a completed offence.

• (3) Once an agent has begun to attempt an offence they have only two options: (i) don’t complete the offence; or (ii) complete the offence.

• (4) Not completing an offence after one has begun attempting it does not have more utility than completing the offence, in fact, it might have slightly less (from 2).

• (5) Therefore, an agent will not be deterred from completing an offence once they have begun to attempt the offence (from 1, 3, and 4).

The idea here is clear enough. If someone wants to murder another person, then, if attempts are punished in the same way as completed crimes, they will have no incentive to refrain from murdering that person if they have started an attempt. Why bother? If their punishment is going to be the same in both instances, why not finish the job? But, so the follow-up argument would go, this is perverse: we should prefer attempts to completed crimes since they are less harmful (even if they do cause some harm). So we shouldn’t allow for attempt liability.

Of course, this argument only really bites on the equivalency version of attempt liability, i.e. the version that holds that attempts should be punished in the exact same way as completed crimes. If a lesser punishment attaches to attempts, then the deterrent effect would return (even if it’s minimal) and you would get the added benefit of deterring attempts as well.

Okay, so that brings us to the end of this series. As you can see, there are some reasonable arguments to make on both sides. I’m not going to try to assess which side wins the debate. In an annoying move, I’m going to leave that up to you.

Attempt Liability and Moral Luck (Part One)

Tweet

(Series Index)

This post is the first in a short series on the combined issues of attempt liability and moral luck. Attempt liability is an idea arising out of criminal law. It holds that a person can be held criminally liable for attempting an offence, such as murder or rape, as well as for completing an offence. And moral luck is….well, moral luck is one of the more interesting conceptual developments in 20th century ethical philosophy. I’ll talk about it in more detail later.

The purpose of this series is to address a simple question: is it right hold someone criminally liable for attempting, as opposed to completing an offence? This entry introduces the basic problems associated with attempt liability, their connection to the problem of moral luck, and some basic arguments in favour of attempt liability. The second entry will consider the arguments against attempt liability.

I’m going to be working off a variety of sources for this post. I’ll provide relevant links as I go along.


1. The Problem(s) of Attempt Liability
So should an attempted murderer be punished in an equivalent manner to a successful murderer? Before you answer that, let’s pump some intuitions with the following case study:

The Poisoning in the Teacher’s Room (A): Mike and Marge both teach at the local high school. They don’t like each other very much, and clash repeatedly during their daily exchanges in the teacher’s room. Finding it’s all too much to take, Mike decides to poison Marge. In the teacher’s room there are two large pots containing loose-leaf tea leaves or ground coffee. Marge drinks coffee every day. So Mike decides to place a quantity of poison in the coffee pot, just prior to her taking some to make her daily cup of coffee (he’ll replace it after, before anyone else takes some poison). He does this and succeeds in poisoning and killing Marge.

The Poisoning in the Teacher’s Room (B): The exact same as above only at the last minute, for some unknown reason, Marge switches from drinking coffee and decides to drink tea instead.

How do you feel about these two cases? Do you think Mike is just as culpable in Case A as he is in Case B? If you do then you are accepting (subject to defeaters) that attempt liability is a legitimate idea. But in doing so you open up the door to some problems. Chief among them being the problem of distinguishing an attempt from a non-attempt.

One of the core values in a liberal society is that people are free to think and act as they wish as long as they do not harm anyone else in the process; one of the hallmarks of the totalitarian society is its attempt to regulate, control and punish thought and behaviour, irrespective of whether it harms anyone else. The problem with allowing for attempt liability is that you may begin the slide from liberalism to totalitarianism. While we may feel comfortable treating Mike from Case A the same as Mike from Case B, in doing so we might set a dangerous precedent for future cases, a precedent that blurs the boundary between an attempt and a non-attempt.

(See the discussion of precedential slippery slope arguments in this post for more on the arguments that might be made here).

Consider the following two variations on the poisoning case:

The Poisoning in the Teacher’s Room C): Mike decides to poison Marge. He reads up about poisoning on the internet, buys some poison and plans out exactly what he is going to do. But at the last minute he gets cold feet and doesn’t put the poison in the coffee pot. Marge lives on oblivious, but Mike later tells a colleague of his plans and they inform the police (this last bit is probably irrelevant from a moral perspective, but important from an evidential one).

The Poisoning in the Teacher’s Room: Mike would love to poison Marge, but he just doesn’t have the courage. He fantasizes about it every day, planning the act meticulously in his mind, but never bringing it to fruition. He tells his therapist, but she sees no threat in his fantasies.

How do we feel about these two cases? I suspect we’d feel that Mike from Case D should be exempt from liability: his idle fantasies are exactly the kind of thing we want to protect from government intervention in a liberal society. But how about Mike from Case C? What he did was clearly less culpable than what he did in case B, but we might still feel like some intervention, or minimal form of liability is merited. Indeed, the criminal law can allow for some liability through the offence of conspiracy. However, by allowing for this we’ve definitely begun to slide from pure liberalism. That might be a perfectly acceptable thing to do — a capacity for nuance and an appreciation for the moral complexity of the real world are qualities we might like the criminal justice system to exemplify — but we have to consider where the boundaries should be drawn, if at all.

Distilling from the preceding discussion, there are perhaps three core boundary-line problems associated with attempt liability:

The Slippery Slope Problem: Should we even begin to punish attempts in addition to completed crimes?

The Equivalence Problem: Should we punish attempts in the exact same manner as completed crimes?

The Gradient Problem: Should we grade attempts in terms of their seriousness, and punish them in accordance with their location on the gradient?

Our primary focus will be on the first two problems, not so much on the third. I want to explore the arguments that propose different answers to those two problems. But before I do that I want to talk a little bit more about the issue of moral luck.


2. The Problem of Moral Luck
The problem of attempt liability links to a more general problem in moral philosophy, namely: the problem of moral luck. This is something which has garnered much attention in the relatively recent past. This largely began in the late 70s/early 80s when Thomas Nagel and Bernard Williams wrote a pair of classic articles on the topic.

The problem of moral luck can be simply stated: what role should luck play in the moral assessment of someone’s actions, accomplishments, failures, achievements and so on? Luck can be roughly defined as any action, event or state of affairs that is outside the control of the person being assessed. In his analysis of the issue, Nagel identified four main varieties of moral luck. They are (taken from the SEP article):

Resultant Luck: This is luck relating to the results of our actions. Poisoning cases A and B give us some idea of what is involved here. In both cases, Mike acted in the exact same way, but the results of his actions were different. In one instance, they led to the death of Marge, and in the other they did not.

Circumstantial Luck: This is luck relating to the circumstances in which one finds oneself. The classic example here being those who found themselves living in Nazi Germany in the 1930s. They likely did things that many of us would do (follow authority, act in their own self-interest) only they did so in appalling circumstances, circumstances which were outside of their control but made them complicit in an atrocity.

Constitutive Luck: This is luck relating to the kind of person that you are. Although we might like to think that we control our character traits and personality tics, our genes and our environment must play a considerable role in determining their content. These factors are beyond our control and could impact upon our moral choices.

Causal Luck: This is luck relating to the antecedent causes of who we are and how we act. Nagel views this as being equivalent to the issue of causation and determinism in the free will debate.

Now I’ll have to be honest and say I don’t see the need to distinguish between constitutive and causal luck. They seem like very similar concepts to me. Nevertheless, I think there is some utility to the resultant/circumstantial distinction. Indeed, within the criminal law — which is where the issue of attempt liability arises — this distinction is significant. This is because crimes are typically distinguished from one another on these kinds of grounds. For example, murder is a result-oriented offence: the actus reus of murder is the death of one person as caused by actions of another. In contrast, rape is a conduct and circumstances-oriented offence: the actus reus of rape is (usually) penetration of a bodily orifice of one person by the penis of another (conduct), without the consent of the person being penetrated (circumstance). So the conceptual and normative issues associated with resultant and circumstantial luck could be important when assessing attempt liability.

I previously said that the problem of moral luck has to do with whether luck should play a role in our moral assessments. That characterisation of the problem is sort of correct, but not quite all the way there. One of the realisations to emerge from the Nagel/Williams exchange in the 70s/80s was that luck does seem to play a substantial role in our everyday moral assessments. That is to say, we already seem to allow for moral blame to attach even in the presence of luck. So, to them and to most contemporary philosophers, the problem of moral luck is not “whether” luck has a role to play but, rather, how to account for the role that it does play. Of course, that’s not to say that the more general normative issue of “whether” is ignored — far from it — but it does suggest an alternative perspective is being taken on the problem.

Anyway, we now need to move away from this problem-setting stage and on to the problem-solving stage. We do this first by looking at three arguments in favour of punishing attempts and treating them as (roughly) equivalent to completed crimes. These arguments are: (i) the control argument; (ii) the harm prevention argument; and (iii) the deterrence argument. Let’s look at each in turn.


3. The Control Argument
The control argument appeals to the most common principle of responsibility, namely: an agent should only be liable for those results (and circumstances) that are within their control. From there it builds a case for treating attempts the same as completed crimes. Using the poisoning cases from earlier as a reference, this is the basic control argument:

• (1) An agent is (only) liable for the results and circumstances that are within their control.

• (2) In terms of results and circumstances, Mike from Case A exercised control over the exact same things (no more, no less) as Mike in Case B.

• (3) Therefore, if Mike from Case A is liable for something, Mike from Case B is liable for the exact same thing.

• (4) Mike from Case A is liable for the murder of Marge.

• (5) Therefore, Mike from Case B is liable for the murder of Marge.

Although the logic here seems valid enough, the conclusion is strange. How can Mike be liable for murder in case B when Marge isn’t dead? After all, murder requires an actual death, doesn’t it? Since the conclusion seems strange we might be inclined to think that at least one of the premises is dodgy, but let’s not be too hasty. I suspect the reason for thinking that the conclusion is strange stems from confusing liability and responsibility. This is something I’ve spoken about before. Roughly, liability is concerned with the price one has to pay for one’s actions; whereas responsibility is concerned with the outcomes that one actually brought about. Since this argument is framed in terms of liability, not responsibility, I think it makes sense: Mike must pay the same price in both cases because he controlled the same things, despite the fact that his actions led to a different result.

Even still, there might be something wrong with the premises. For instance, we might argue that control is not the only thing relevant to liability, that utilitarian principles can also be used to determine who should pay the price for something. In that case, premise (1) would be faulty. But that wouldn’t necessarily spell the end for attempt liability because utilitarian principles might just as easily be used to support the case for attempt liability. Indeed, this is exactly what the next two arguments hold.


4. The Harm Prevention Argument
Like the control argument, the harm prevention argument works from a very simple idea. The idea is that the purpose of the criminal law is to identify those wrongs which we would prefer not to occur. And since those wrongs are usually (and probably preferably) linked to harms, it would seem that the criminal law is designed to prevent harm. Think about it like this. The reason we classify murder as a crime is because we don’t want people to kill one another. And the reason we do this is because deaths are harmful and we wish to prevent harm.

But then, if our goal is harm prevention, why should we wait until the harm has been caused before intervening? In other words, why doesn’t the following argument hold?

• (1) The criminal justice system ought to prevent harm.

• (2) Intervening before a crime has been completed (but after it has been attempted) prevents more harm than intervening after the crime has been completed.

• (3) Therefore, the criminal justice system ought to be willing to intervene before a crime has been completed, not just after.

This argument is fine, in so far as it goes. The problem is that it doesn’t go far enough. While it might be true that, if we’re interested in harm prevention, we ought to try to prevent harm and not just step in after it occurs, this doesn’t say anything about whether we should hold someone liable for attempting a crime. In other words, the argument fails to answer the question: why can’t we just prevent the crime and leave it at that (without punishing the attempt)?

There are a number of possible replies (impracticality, epistemic hurdles etc). The next argument is one of them.


5. The Deterrence Argument
Let’s say we accept the basic tenets of the harm prevention argument. What we then need is some principle to plug the gap between intervention and liability. A deterrence argument might be exactly what we need. A deterrence argument will work off the idea that there are certain incentives that make people more likely to respond or behave in a particular way in the future. In many ways, the goal of any social engineer is to craft a network of incentives that encourages people to behave in ways you like, and deters them from behaving in ways you do not like.

When it comes to attempt liability, the proponent of deterrence is going to argue that intervention+punishment is going to be a more effective deterrent than intervention on its own. And if it is a more effective deterrent, then it is going to prevent more harm going into the future than an intervention.

This leads us to the following argument:

• (1) The criminal justice system ought to prevent as much harm as possible.

• (2) Intervening and punishing attempts (i.e. creating a system of attempt liability) will prevent harm than just intervening before crimes are completed (because it provides a greater deterrent).

• (3) Therefore, the criminal justice system ought to create a system of attempt liability.

This argument, which is subtly different from the harm prevention argument, provides some justification for attempt liability. The second premise would be supported by the deterrence-based reasoning that I outlined in the two preceding paragraphs.

Is the argument any good? Well, note the change in premise (1) from the version in the harm prevention argument. One might argue that this change is both crucial and problematic. It is crucial because without it the deterrence based objection to intervention without liability won’t work. It is problematic because it may claim too much for the criminal justice system. Should we really aim to prevent as much harm as possible? Probably not, especially if doing so will also prevent us from doing other things that we might value. Whether deterrence actually do that is a question worth pursuing. All I’ll say here is that by using deterrence as the justification for imposing liability, we may slide down the slope towards totalitarianism. After all, totalitarian societies might be very safe places to live, but they achieve this at the expense of other values.

Okay, let’s leave it there for now. In part two, we’ll look at the arguments on the other side of the debate.

Free Will, Punishment and Responsibility (Series Index)

Tweet

I've been blogging a lot recently on the philosophy of responsibility, punishment and, to a lesser extent, crime. Since I'm likely to continue addressing those topics in the coming weeks and months, I thought it was time to provide an index to all the posts I've written so far. I've divided them into three specific groups, but the divisions are far from pure. Anyway, you can expect this to grow somewhat in the near future.


1. Free Will and Moral Responsibility

• Free Will: a Conceptual Framework
• Who's Afraid of Determinism (Part One, Part Two, Part Three)
• Action, Responsibility and the Ability to do Otherwise (Part One, Part Two)
• Present Luck and Little Agents
• Fischer, Frankfurt and Flickers of Freedom (Part One, Part Two)
• The Frankfurt Counterexample Defended by Nucci
• List on Free Will and Determinism (Part One, Part Two)
• Waller's Argument Against Moral Responsibility (Part One, Part Two)
• Are hierarchical theories of free will and responsibility plausible?
• Freedom: Non-frustration, Non-interference and Non-domination (Part One, Part Two)
• Pereboom's Case for Hard Incompatibilism
• Did my brain make me do it? (Part one, Part Two)
• Psychopaths and Moral Responsibility: Empirical and Philosophical Issues

2. Theories of Punishment

• The Ethics of the Death Penalty
• Moore on Justifying Retributivism (Part One, Part Two)
• Theories of Desert
• Brooks on Retributivist Arguments Against the Death Penalty (Part One, Part Two)
• The Problem of Punishing the Innocent
• The Ethics of Pre-Punishment (Part One, Part Two, Part Three)
• Kramer on the Purgative Rationale for Capital Punishment (Part One, Part Two)
• Consequentialist Theories of Punishment (Part One, Part Two, Part Three)
• Is the death penalty irrevocable? (Part One, Part Two)
• Can death be a fitting punishment?
• Does the criminal law deter? (Part One, Part Two)
• Are we innate retributivists? Review of the Psychological Evidence
• The Psychology of Revenge: Biology, Evolution and Culture

3. Criminal Responsibility and Liability

• Madness, Badness and Neuroimaging-based Responsibility Assessments (Part One, Part Two, Part Three)
• Vincent's Structured Taxonomy of Responsibility Concepts
• Taking Responsibility vs. Being Morally Responsible
• Vincent on the Responsibility-Liability Gap
• Attempt Liability and Moral Luck (Part One, Part Two)
• Attempted Crimes and Changes of Mind
• Addiction and Responsibility (Part One, Part Two, Part Three)

The Problem of Punishing the Innocent

Tweet

This post looks at a classic objection to utilitarian theories of punishment. It starts by setting out the objection as clearly as possible. It then considers the potential replies to the objection.



1. The Problem of Punishing the Innocent
A utilitarian theory of punishment maintains that it is right and proper to punish people for committing a crime, if doing so serves one (or more) of a set of morally desirable goals. According to Howard Simmons’s book Moral Desert: A Critique, these goals can include the following: (i) preventing repeat offences (through incapacitation); (ii) encouraging deterrence (through signalling an intent to punish); (iii) promoting moral education (by forcefully communicating normative standards); (iv) providing emotional catharsis (for the victim or family of the victim); and (v) rehabilitating the offender.

Some of these goals may be more commendable than others (Simmons certainly thinks so), and aiming to achieve one may prevent or undermine the achievement of others. These are things that any fully-developed utilitarian theory of punishment will need to work out. But for the purposes of this post I’ll be ignoring them. Instead, I’ll be focusing on a problem that is supposed to confront nearly all utilitarian theories of punishment (with perhaps the sole exception of pure rehabilitative theories). This is the problem of innocent victims.

The problem can be put like this:

• (1) If utilitarian theories of punishment are true, it is right and proper to punish people if doing so will serve one (or more) of the following goals: (i) preventing repeat offences; (ii) deterrence; (iii) moral education; or (iv) emotional catharsis for victims of crime.

• (2) It is possible that punishing innocent people could help us to achieve goals (i)-(iv).

• (3) Therefore, if utilitarian theories of punishment are true, it could be right and proper to punish the innocent.

• (4) It is not right and proper to punish the innocent.

• (5) Therefore, utilitarian theories of punishment must be false.

In the remainder of this post, I want to see whether this problem can be dealt with by proponents of utilitarianism. Before doing so, I want to briefly check to see whether the above argument is prima facie plausible.

Looking first to premise (1), I note that this is simply a restatement of the utilitarian thesis. As such, I can see no reason to reject it. Turning then to premise (2), I appeal to the standard thought experiments that are used to support it.

Sheriff Case: Imagine that you are the sheriff in an unruly town that has recently been struck by a series of murders. The townspeople are convinced that a certain person (call them “Jack”) committed the murder. They call upon you to execute Jack in your official capacity. But you are pretty sure that Jack is innocent. You try to tell the townspeople your reasons for thinking this, but they don't seem to care. You know that if you do not carry out their wishes, the town is likely to descend into anarchy and violence. These outcomes are contrary to the goals you seek to achieve by having a system of punishment in the first place.

The suggestion is that, if you were a utilitarian, you should have no objection to executing Jack in this scenario. This seems, prima facie, a plausible claim to make.

Premise (3) follows from the conjunction of (1) and (2), so there can be no objection to it here. That leaves us with (4) as the one potential remaining source of controversy. Note here that (4) is intended to be a description of an intuitive judgment about punishing the innocent. Of course, we are entitled to question how seriously we should take such intuitive judgments. But this kind of questioning can only come after we engage in a more detailed analysis of the assumptions underlying the above argument. Looking at the surface level, we have no reason to reject it.

Since (5) would follow from (3) and (4), it looks like the argument is indeed prima facie plausible. The question we must now ask is whether, once we go beyond surface appearances, the argument remains plausible.



2. Possible Replies
So how might a utilitarian undercut the plausibility of the preceding argument? Two strategies — hinted at above — seem workable: (i) challenge premise (2) by arguing that punishing the innocent would never lead to the desirable consequences listed above; or (ii) challenge premise (4) by arguing that punishing the innocent is not necessarily wrong or that the concept of innocence is irrelevant to utilitarian justifications of punishment. Let’s consider both strategies in turn and, to make things slightly less complex, let’s assume that deterrence is the consequence we are trying to achieve with our system of punishment.

Granting that assumption, proponents of the first strategy might argue in the following manner. First, they might argue that deterrence — at least in the present context — requires the presence of an act of wrongdoing. In other words, they might argue that since our goal is to deter people from wrongdoing, we must assume that there are actual wrongdoers out there. Second, they might argue that in order for a punishment to have a deterrent effect, there must be a link between it and the act (or acts) of wrongdoing that are being deterred. That is to say, the punishment must only arise after acts of wrongdoing, not after other kinds of act.

Think of it in terms of the conditioning experiments of the behaviourist psychologists: in order for a particular behavioural response to be provoked (or suppressed) by a particular kind stimulus, there must be a reliable link set-up between the presence of the stimulus and the presence (or absence) of the behaviour. If there is no reliable link between the two, then the chances of provoking (or suppressing) the relevant behaviour are limited. So, similarly, if the link between wrongdoing and punishment is broken, we can’t expect the system to deter people from wrongdoing. This is what would happen if we started to punish the innocent.

We can put this objection like this:

• (6) The goal of a utilitarian system of punishment is to deter future acts of wrongdoing.

• (7) In order for a particular kind of behavioural response to be successfully provoked (or suppressed) by a particular kind of stimulus, there must be a reliable link between the response and the stimulus (i.e. if one happens then the other must also happen either at the same time or soon after).

• (8) A system of punishment is a kind of stimulus and an act of wrongdoing is a kind of behavioural response.

• (9) Therefore, in order for a utilitarian system of punishment to be successful, there must be a reliable link between the punishment and acts of wrongdoing.

• (10) If we start to punish the innocent, then there will be no reliable link between punishment and acts of wrongdoing.

• (11) Therefore, a successful utilitarian system of punishment cannot allow for the innocent to be punishment.

There are two objections to this argument. First, there is an obvious objection to be made to the theory of human psychology appealed to in premise (7). Even if we stay within the behaviourist tradition, we find those who object to the idea that the link between a stimulus and a response must be 100% consistent. Thus, for instance, B.F. Skinner argued that inconsistent (or, rather, probabilistic) “schedules of reinforcement” could be even more effective (addictive, in fact) for conditioning certain kinds of behaviour. If that’s right, then a system of punishment which is not 100% consistent in who it punishes might be better than one that is. I'm not an expert in behaviourist psychology so I could well be wrong about this, but it seems like a possible objection nevertheless.

The second objection is not an objection to the argument per se, but, rather, an objection to its relevance to the problem of punishing the innocent. Recall that premise (2) of the original argument merely appeals to the “possibility” that punishing the innocent might serve a deterrent goal. This mere possibility is then thought to be enough to counter utilitarianism. In other words, premise (2) is suggesting that if the utilitarian theory is not true in all possible worlds, then it is not an acceptable theory. (This focus on truth across all possible worlds is actually a common argumentative ploy in moral philosophy). Premise (2) does not make any claims about what is needed for a successful system of utilitarian punishment. So, in its present form, premise (11) does not actually defeat premise (2).

This raises a general concern about the use of logical possibility as a standard for testing ethical arguments. Is it really true that we need to prove that our ethical principles are true in all possible worlds before we can accept them? I was at a seminar just last week where one participant defended the use of this standard in ethics. At the time I didn’t voice any objections, but on reflection I worry that it sets the bar at a ridiculously high level. Admittedly, that high level is common to many philosophical disciplines (epistemology anyone?), but surely ethics needs to be more sensitive to both our inability to explore all possible worlds and to the outlandish nature of some possibilities? Surely it would be better use probability-of-happening-in-the-actual-world as a standard?

I offer these as questions for the time being since I don’t know exactly how to answer them. (Does anyone know of good papers addressing the specific issue of standards of success for ethical theories?)


3. Could it be right to punish the innocent?
We turn now to consider the second strategy in responding to the original argument. This strategy, as noted above, focuses on the plausibility of premise (4). On the face of it, this premise looks pretty robust: the idea of punishing the innocent seems about as clear an instance of a wrong as one could imagine. Thus, this seems like a less promising strategy for the utilitarian. Still, there are a couple of things that can be said in its defence.

For one thing, there is the whole idea that the appeal to “innocence” is question-begging. Why? Because it is a holdover from a retributive, desert-based theory of punishment. It has no place in a utilitarian theory. After all, the idea that some people are innocent and so shouldn’t be punished is clearly based on the notion that responsibility and desert are both coherent and significant moral concepts. It is open to a utilitarian to reject both of these notions.

They could do so by appealing to some form of hard determinism. According to hard determinism, no one is ever responsible for anything they do, and since responsibility usually goes hand-in-hand with desert, it would follow that no one ever deserves punishment. This, however, does not mean that punishment is unjustified. It could be that punishment is a kind necessary evil, something that is justified because it is needed to maintain a desirable level of social order. This is, in many ways, consistent with the traditional utilitarian approach to punishment: all harm is evil and punishment is a form of harm, but some evil is acceptable if it allows us to achieve a greater good.

Now, to be fair, hard determinism has plenty of critics: it has a number of counter-intuitive and unwelcoming implications. If you wished to use it here, you’d need to know exactly how hot the philosophical water into which you were getting was. Still, it is a possibility for the defender of utilitarianism.

Another objection to (4), this one made by Simmons in the course of defending his own utilitarian theory, focuses on the alleged unjustifiability of punishing the innocent in the extreme one-off cases. It’s these kinds of cases — such as the Sheriff one — that are most commonly used to support the problem of punishing the innocent. What’s more, it’s these kinds of cases that are not vulnerable to the “punishing the innocent is not likely to be successful”-argument that was discussed above.

But when looking at these extreme cases, Simmons argues that the intuitive pull of the “punishing the innocent is wrong”-judgment is much less robust. These are hard cases (or dilemmas), ones in which both outcomes seem undesirable and so intuitions clash. He thinks that if we really thought hard about a case like the Sheriff one, and if we accepted all the assumptions it asked us to make about the probability of certain outcomes, we might be less inclined to reject the utilitarian approach. Thus, premise (4) is at least questionable in these extreme cases.

Note that this objection shows how the sword of logical possibility cuts both ways: if the utilitarian must abandon their position if a logically possible, if improbable, scenario suggests that punishing the innocent might achieve desired outcomes, so too must the non-utilitarian abandon their intuitive judgment if a logically possible, if improbable, scenario leads us to seriously doubt our previously robust moral intuitions. Perhaps this is yet another reason to abandon the logical possibility standard.

Moore on Justifying Retributivism (Part Two)

Tweet

(Part One)

This is the second post in a brief series looking at Michael Moore’s article “Justifying Retributivism”. In part one, I introduced the conceptual framework Moore uses to support his justification of retributivism; in this second part, I turn to the actual argument Moore uses to justify retributivism.

Before we start, it’s worth reemphasising what it is that Moore is really trying to justify with his argument. As we saw the last day, Moore takes retributivism to be the principle that we have a moral obligation to punish those, and only those, who are guilty of wrongdoing. Moore interprets this principle in a deontological manner. In other words, he believes that the obligation to punish holds irrespective of any positive or negative effects that the punishment might have. This is partly because he takes punishment to be an intrinsic good. In trying to justify retributivism, Moore is trying to justify this complex of views about the nature of punishment.

How does he do it? That is what we are about to see.


1. The Thought Experiment and the Argument
Moore takes our intuitive judgments about particular cases seriously. He thinks we derive our more general ethical principles via abductive inference from particular judgments. This results in a kind of reflective equilibrium in which our general principles cohere with (are in equilibrium with) our intuitions.

Aiming for reflective equilibrium is probably the most common strategy in applied ethics, mainly because, as David Boonin points out, no other methodology seems to be available. The differences, such as they are, come between those who afford decisive weight to our intuitions and those who afford decisive weight to rationally compelling general principles. Peter Singer would be an example of someone who belongs in the latter camp; Moore seems to belong in the former.

This obviously influences how Moore argues for retributivism. He does so by first developing a thought experiment, asking us to consider our judgments about this thought experiment, and then arguing that only retributivism can account for these judgments. This is an acceptable form of argument, provided one accepts the methodological strategy outlined above.

Let’s look now at the thought experiment Moore uses as the basis for his argument. It is inspired by the story of the Russian nobleman in The Brothers Karamazov . Roughly, the story is the following: there is nobleman who owns a particularly nasty and vicious pack of dogs. On one occasion he commands these dogs to tear a young child limb from limb, and he forces the child’s mother to watch this barbaric act. The question arises: what do we think is the appropriate response to this state of affairs? The suggested answer is: we think the nobleman should be punished.

This suggested answer provides some prima facie support to retributivism, but Moore knows he isn’t home and dry yet (far from it) since alternative theories of punishment might sustain our intuitive judgment about this particular case. So Moore adds a couple of qualifications.

First, we are asked to assume that no purpose other than that of harming the guilty party can be served by the punishment. In other words, we are to assume that the punishment could not serve as a deterrent for would-be wrongdoers - perhaps this is because we are part of an island society that is about to disband or collapse.

Second, we are to consider two variations on the thought experiment: (i) the first-person version in which we ourselves are the Russian nobleman and (ii) the third-person version where we simply bear witness to the Russian nobleman’s act.

What happens when we include these qualifications? Moore thinks we will still think that punishment is warranted. We will think it warranted in the first-person scenario due to our strong sense of moral guilt for what we have done; and we will think it warranted in the third-person scenario due to an appropriate feeling of empathy for the victims. This suggests that our intuitive judgments are strongly in line with the retributivist principle. Furthermore, since no other moral purpose can be served by punishment in this scenario, we are inclined towards the view that punishment is an intrinsic good.

All of this can be crafted into a more formal argument in favour of retributivism. As follows:

(1) We are justified in accepting whatever moral principles and concepts best account for our intuitive judgments about particular cases. 

(2) The principle of retribution and the intrinsic goodness of punishing the guilty best account for our reaction to the two scenarios outlined above. 

(3) Therefore, we are justified in accepting the principle of retribution and believing in the intrinsic goodness of punishing the guilty.

The diagram below summarises most of the above.

2. Charges and Responses
Moore acknowledged that his argument suffers from a number of apparent defects. Five of which have featured in the literature and each of which he attempts to respond to. Some of these criticisms get into larger debates in moral philosophy, but I’ll try my best to summarise them here.

(a)  The Charge of Circularity : A common rejoinder to Moore’s argument is to charge it with circularity. It would appear that his general retributive principle is being sustained by nothing more than a retributive judgment in a particular case. This seems to beg the question as to whether retributivism is justified. Moore notes that there a few different ways of making the charge of circularity, not all of which are compelling. For instance, he thinks it is clearly wrong to think that a judgment about a particular case is literally the same thing as the principle one generalises from that particular case. An alternative way of making the charge is to argue that since the judgment in the particular case is about the desert of the guilty it makes the general principle about the desert of the guilty trivial. Moore rejects this. He argues that the judgment in the particular case is not about desert but, rather, about punishment simpliciter . Desert is the more general concept that helps to explain the judgment about the particular case. 

(b) The Charge of Illegitimacy : Some argue that Moore’s methodology — that of deriving general principles from judgments about particular cases — is illegitimate, that you can only derive principles of this sort from other, more basic, principles. Moore responds by saying that such a charge forces us back into the kind of foundationalist epistemology that has largely been rejected. Now, there are certainly some cogent contemporary defenders of foundationalism, but I think Moore has a reasonably defensible point here. 

(c) The Charge of Subjectivity : Another methodological criticism is to make the point that Moore’s coherentist justification of general ethical principles cannot help us to arrive a objective ethical truths. At best, it can help us to arrive at a subjectively coherent web of beliefs. This gets a the issue of how much weight is to be attached to intuitive judgments in ethical reasoning, an issue I’ve covered in my series on evolutionary debunking arguments. As noted in that series, how seriously this charge is taken depends on one’s metaethical views. An anti-realist is likely to be unperturbed by the reliance on subjective judgments, but a realist might be. 

(d) The Charge of Emotionality : It could be argued that the responses to the particular cases outlined engage the non-rational, emotional faculties — indeed, Moore encourages such an interpretation when he appeals to feelings of guilt and the sense of empathy — and that they cannot then be used to support rational moral principles. This criticism doesn’t take issue with the particular sort of emotions being tweaked by the thought experiments, but rather with the more general use of emotions. This criticism only works if one thinks that the emotions have no cognitive element to them, something which many would now reject. It is also worth bearing in mind that there are some who argue that moral judgments are, properly understood, entirely constituted by emotions. Moore rejects this view so he is inclined to downplay the strength of the connection between the emotions and the moral principles that we derive from the particular cases. He argues that emotions are our main heuristic guide for discovering moral truths, but nothing more than that. 

(e) The Charge of Narrowness : A final criticism of Moore’s argument is that the retributive principle derived from our responses to particular cases like those outlined above cannot be general enough to cover all instances of culpable wrongdoing; it can only cover instances of wrongdoing that are sufficiently similar to those outlined above. This is a problem for Moore since he thinks that the retributive principle undergirds the entirety of the criminal justice system. The response here is complex since it depends on how “wrong” one takes all criminal acts to be. It is, after all, possible that the legal system is too harsh or even outright immoral in criminalising certain forms of conduct. Moore’s approach here is piecemeal. In the case of clear moral wrongs that are also criminal, he thinks the principle will carry over, although, understandably, the strength of the deserved punishment will decrease with the gravity of the wrongdoing. As for acts that are morally neutral or morally good that happen to be criminalised, he thinks the principle will not carry over. So much the worse for any legal system that engages in such arbitrary criminalisation.

This then completes Moore’s defence of retributivism. No doubt the methodology will still seem questionable to many and the conclusion raises further questions: how do we know when someone deserves punishment? Which forms of punishment are proportionate responses to which forms of wrongdoing? And so on. I’ll be looking at some of these issues in the near future.

Moore on Justifying Retributivism (Part One)

Tweet

Well, here I am. After a couple of months hiatus (my longest ever) I am now back and ready to blog. As mentioned yesterday, I'm going to be integrating the content of this blog more directly with my ongoing research and teaching. This means you can expect to find a lot more posts dealing with applied ethics, the philosophy of responsibility, the philosophy of evidence, and game theory. Don’t worry though, I’m hoping to do some posts on the philosophy of religion when I get a chance (maybe I’m being naive).

Anyway, for my first set of substantive posts, I’m going to be looking at retributive justice and the death penalty. This is part of an attempt to prepare for some lectures I’ll be doing later in the year on the ethics of the death penalty. I was going to launch directly into a discussion of the arguments for and against the death penalty (once again using the book Contemporary Debates in Applied Ethics as my guide), but since those arguments rely, in part, on an understanding of retributive justice, I thought I’d back-up and consider that concept in a little more depth.

As a result, I’m going to spend the rest of this post focusing on a paper by the legal philosopher Michael Moore (not to be confused with the documentary film director of the same name). The paper is the following:

• “Justifying Retributivism” (1993) 27 Israel Law Review 15

The title hints at Moore’s own philosophical leanings (he has been one of the most prominent defenders of retributivism in the recent past) but in the article he does a good job clarifying the nature of retributivism, and presenting arguments that can be offered in its defence. I’m going to look at both of these in what follows.


1. What is Retributivism?
We can begin with a simple definition. As Moore puts it, “retributivism is the view that we ought to punish offenders because and only because they deserve punishment.” (p. 15) In addition to this, retributivism is usually taken to include a proportionality constraint. In other words, the retributivist usually believes that the punishment of the guilty must be proportionate to their level of wrongdoing. Since I’m ultimately going to be considering the death penalty, it’s worth flagging at this point that death is taken by some to be the proportionate punishment for murder and perhaps some other offences.

Moore thinks that there are a number of ambiguities present in this basic definition. These need to be addressed and he spends several pages trying to do so. We can summarise the results of this exercise as follows:

• The definition of retributivism appeals to the concept of desert ( i.e. the idea that the connection between an agent and his act sometimes deserves a certain kind of treatment). All theories of punishment, retributive and non-retributive, make some sort of appeal to this concept. What distinguishes the retributivist is his belief that desert is both a necessary and sufficient condition for punishment. What’s more, the retributivist also typically holds that desert is an exclusive condition for punishment, i.e. no other condition will warrant punishment. Punishment may well have as a consequence the deterrence of wrongdoing or the increase of the aggregate sum of happiness in a society, but these are happy accidents, not alternative grounds for punishment.

• Like all moral concepts, retributivism, if accepted, places moral demands on both state officials and citizens. But what is the nature of these demands? Sometimes it is believed that retributivism merely makes punishment of the guilty morally permissible - something we can do if we wish, but which are not obliged to do. Moore thinks this is too weak. He thinks that retributivism actually makes punishment of the guilty morally obligatory. In other words, it gives us a duty to punish the guilty.

• A popular distinction — one endorsed by John Rawls and HLA Hart — is that between the justification for particular acts of punishment and the justification for institutions of punishment as a whole. Hart, for example, held that particular acts of punishment could be justified on retributive grounds, while institutions of punishment could be justified on more utilitarian grounds. Moore doesn’t buy this distinction, he thinks the justification for both must be the same. He has a reason for this that will become apparent when we consider his arguments in favour of retributivism.

• Retributivism is often thought to be justified solely on deontological grounds. But Moore thinks this is need not be the case. The retributivist believes that punishment is an intrinsic good (one among many possible intrinsic goods that need to be promoted) and so it is conceivable that he could justify punishment on consequentialist grounds. Nevertheless, Moore considers himself to be a deontologist about punishment and this comes through in his subsequent arguments.

With these clarifications out of the way, we can proceed to consider how one goes about arguing in favour of retributivism.


2. Arguing for Retributivism
Moore identifies three possible ways in which the retributivist can argue for his position:

(a) The Conceptual Argument : The retributivist can argue that his is the only acceptable theory of punishment because the concept of punishment, properly understood, can only be applied to those instances in which people are harmed because they deserve it. This argument would involve a typical exercise in conceptual analysis. Such an analysis is unlikely to be persuasive since even if it is correct it leaves open the deeper question of whether punishment is something we wish to have at all. 

(b) The Functional/Interpretive Argument : The retributivist can argue that his theory offers the best possible explanation (or “interpretation”) for the institutions of punishment that we currently happen to have. This argument obviously suffers from the same flaw as the conceptual argument, viz. it leaves open the question of whether we should have these institutions in the first place. 

(c) The Moral Argument : The retributivist can argue that his theory tells us what the morally correct reponse to certain acts of wrongdoing actually is. This kind of argument will appeal to whatever criteria or standards we take to justify moral beliefs. So if we are consequentialists, the argument will involve a cost-benefit analysis pointing to the respective goods that can be achieved by only punishing those who deserve to be punished; alternatively, if we are non-consequentialists, we appeal to whatever standards we use to justify actions as being intrinsically right.

Moore is, perhaps unsurprisingly, solely concerned with the third style of argument and his goal is to provide a version of it. This raises the obvious question: what standards will he appeal to when justifying the moral propriety of retributivism? At this point in the article, Moore launches into a discussion of the distinction between intrinsic and instrumental goods, on the one hand, and first- and second-order principles in ethics, on the other. It’s pretty interesting stuff, but I’m going to skip over much of it.

The basic idea is that the intrinsic/instrumental distinction is a metaphysical one concerning different types of value, whereas the first-order/second-order distinction is an epistemic one concerning the way in which we come to know of ethical truths. Moore holds that first-order principles are reached via an abductive inference from more particular second- or indeed third-order principles arising from particular cases. In other words, he is a kind of ethical intuitionist, in that he assigns considerable weight to intuitive judgments about particular cases, and uses these judgments to develop more general ethical principles. Frances Kamm is one of the more vigorous proponents of this approach to ethics.

This obviously has implications for the overall argument offered in favour of retributivism. Moore thinks that retributivism is a first-order principle, describing an intrinsic good, that we are categorically obliged to honour in our actions, even if this means we fail to maximise the overall good. Given his intuitionistic predilections, he justifies this conclusion by first considering particular cases, then deriving conclusions about intrinsic good and first-order principles from our judgments about these cases, and then showing why the principle arrived at cannot be followed in a consequentialist manner. I’ll consider this argument the next day.